As you can see, the long-term trend is indeed an improvement in attitude of about 1% per year. Unfortunately, there's also a big disconnect for a couple of years right around 2004, when it became a major political issue after the Massachussetts Supreme Court decision. What this (might) mean is that although long-term trends are up, an advertising blitz doesn't affect both sides equally. For a short period it energizes the anti-gay forces more than the pro-gay forces. This would, of course, be bad news for California's gay marriage initiative this November.
Franklin isn't sure if we'll see the same effect this year that we saw in 2004, and neither am I. Obviously I hope we don't, but this is going to be one close contest.